🚀 go-pugleaf

From article <1991Jan05.222639.6387@dirtydog.ima.isc.com>, by karl@ima.isc.com (Karl Heuer):
> In article <1991Jan5.182428.615@mathrt0.math.chalmers.se> augustss@cs.chalmers.se (Lennart Augustsson) writes:
>>[In C extended with anonymous functions, we could write compose():]
>>	typedef int (*intintfun)(int);
>>	intintfun compose(intintfun f, intintfun g) {
>>	    return (int ANONYMOUS(int x) { return f(g(x))) };
>>	}
>>...the point is that you cannot add unnamed functions in a natural way
>>without getting the additional semantic power.
> 
> I disagree.  Since the return value of the function references objects that go
> out of scope before the value would be used, I would expect this to analogous
> to   int *h(int i) { return &i; }   , which is undefined.

The example is correct. The construct 

    int ANONYMOUS(int x) { return f(g(x)) }        [1]

creates a new function value that has the _values_ of f and g bound
up in its definition. Since the _values_ of f and g are used, scope
plays no part in the value created by [1]. Remember, names have scope--
values do not.

Here is an example that might help:

    int foo (int a, int b) {
        return a + b;
    }

Do you say 'The return value of foo() references objects that go out
of scope before the value is used.'?  Of course not. The expression 
a + b creates a new value; this new value has no reference to a or b.

In the same way, the expression [1] creates a new value; this new value
has no reference to f or g. (Of course, the _values_ of f and g are 
used to create the _new_ value.)

This is really much easier to think about in a functional language--
especially one that allows curried functions and partial application. 
-- 
Norman Graham

<norman@a.cs.okstate.edu>                 Standard Disclaimer Applies
{cbosgd,rutgers}!okstate!norman